Genomic analysis of gene expression relationships in transcriptional regulatory networks
|Haiyuan Yu*, Nicholas Luscombe*, Jiang Qian and Mark Gerstein|
Abstract From merging a number of data sources, we created an extensive map of the transcriptional regulatory network in yeast, comprising 7419 interactions connecting 180 transcription factors (TFs) with their target genes. We integrated this network with gene-expression data, relating the expression profiles of TFs and target genes. We found that genes targeted by the same TF tend to be co-expressed, with the degree of co-expression increasing if genes share more than one TF. Moreover, shared targets of a TF tend to have similar cellular functions. In contrast, the expression relationships between the TFs and their targets are much more complicated, often exhibiting time-shifted or inverted behavior. (Full Text)
Supplementary Figure 1. "Three examples showing simultaneous (A), time-delayed (B), and inverted (C) relationships in the expression profiles. Note there are only 8 time points for each profile, while in the real yeast cell-cycle data there are 17 time points. Also, the expression ratio is not normalized, whereas in the real data each profile is normalized so that the averaged expression ratio is 0 and the standard deviation is 1. The thick segments of the expression profiles are the matched part. (D) The corresponding matrix E for the expression profile shown in (A). The corresponding matrix D is not shown because in this case the match score (the maximal score) is from E and not D. The numbers outside the border of the matrix are the expression ratio shown in (A). The black cell contains the overall match score S for these two expression profiles, and the light gray cells indicate the path of the optimal alignment between the expression profiles. The path starts from the match score and ends at the first encountered 0. (E) The corresponding matrix E for the expression profile shown in (B). Note the time-shifted relationship and how the length of the overall alignment can be shorter than 8 positions. (F) The corresponding matrix D for the expression profiles shown in (C). The matrix E is not shown because the best match score is not from this matrix in this case."
Supplementary document: contains the supplementary tables and all supllementary metarials that the manuscript refers to.
| ||1. Yeast regulatory networks|
| ||2. Single input motifs|
| ||3. Multi-input motifs (regulators)|
| ||4. Multi-input motifs (targets)|
| ||5. FeedForward Loops|
| ||6. Expression relationships for Cho et al's dataset|
|* Joint first-author|
|Last modified on June 13th, 2003|